Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Right fit Database Technology for Discounts

Sinander, Annelie LU and Galonja, Alexandra LU (2022) EITL05 20221
Department of Electrical and Information Technology
Abstract (Swedish)
Detta examensarbete har utförts i syfte att jämföra hur fyra olika databashanterare (DBMS) presterar inom ett begränsat antal områden för att därigenom dra slutsatser kring vilken som fungerar bäst för att lagra data om rabatter på IKEA. De databashanterare som testas är MongoDB, MariaDB, Neo4j och PostgreSQL. För att åstadkomma en jämförelse utfördes en form av testning - en benchmark. Benchmarken består av att en databas sätts upp för varje DBMS som ska jämföras. Därefter lagras rabattdata i respektive databas och exekveringstiden mäts för varje testdatabas. För att ackumulera information kring hur databasen sätts upp men också kunskap om rabatterna som finns hos IKEA, utfördes två intervjuer där denna information framkom.

Resultaten... (More)
Detta examensarbete har utförts i syfte att jämföra hur fyra olika databashanterare (DBMS) presterar inom ett begränsat antal områden för att därigenom dra slutsatser kring vilken som fungerar bäst för att lagra data om rabatter på IKEA. De databashanterare som testas är MongoDB, MariaDB, Neo4j och PostgreSQL. För att åstadkomma en jämförelse utfördes en form av testning - en benchmark. Benchmarken består av att en databas sätts upp för varje DBMS som ska jämföras. Därefter lagras rabattdata i respektive databas och exekveringstiden mäts för varje testdatabas. För att ackumulera information kring hur databasen sätts upp men också kunskap om rabatterna som finns hos IKEA, utfördes två intervjuer där denna information framkom.

Resultaten av testningen och intervjuerna visade att PostgreSQL, den databashanterare som används idag, presterade bäst överlag. Postgres uppvisade lägst latens och är den enda av de testade databashanterarna som inte medför några licenskostnader. Resultaten från en av databashanterarna, MongoDB, bestod endast av uppskattningar och inte exakta mätningar och därför är inte slutsatsen helt entydig. Fortsättningsvis visade det sig att Neo4j hade väldigt avvikande värden jämfört med resterande databashanterare, vilket kan ha berott på en ofördelaktigt konfigurerad databas.

Slutligen presenteras ett antal utvecklingsmöjligheter som visar på att det finns en stor potential att vidare utforska frågan kring hur rabattdatan ska lagras. IKEA föreslås emellertid att även fortsättningsvis använda den klara vinnaren PostgreSQL som databashanterare. (Less)
Abstract
This thesis has been conducted in order to compare how four different database management systems (DBMSs) perform in a limited number of areas, in order to draw conclusions about which one works best for storing discount data at IKEA. The DBMSs being evaluated are MongoDB, MariaDB, Neo4j and PostgreSQL. To enable a comparison, a form of testing was performed - a benchmark. The benchmarking consists of setting up a database for each DBMS to be compared. Discount data is then stored in each database and the execution time is measured for each test database. To accumulate information about how the database is set up but also about the different discounts used by IKEA, two interviews were conducted where this information was able to emerge.

... (More)
This thesis has been conducted in order to compare how four different database management systems (DBMSs) perform in a limited number of areas, in order to draw conclusions about which one works best for storing discount data at IKEA. The DBMSs being evaluated are MongoDB, MariaDB, Neo4j and PostgreSQL. To enable a comparison, a form of testing was performed - a benchmark. The benchmarking consists of setting up a database for each DBMS to be compared. Discount data is then stored in each database and the execution time is measured for each test database. To accumulate information about how the database is set up but also about the different discounts used by IKEA, two interviews were conducted where this information was able to emerge.

The results from the testing and the interviews showed that PostgreSQL, which is the DBMS used today, performed the best overall. This was the case for the licensing metric, for instance, where Postgres was the only DBMS that does not come with a licensing cost. The results from one of the DBMSs, MongoDB, only consisted of estimates and not exact measurements and therefore the conclusion is not completely unambiguous. Furthermore, Neo4j's results when it came to the latency and scalability were deviating compared to the remaining DBMSs, which may have been due to an disadvantageously configured database.

Lastly, a number of development opportunities are presented which show that there is an immense potential to further explore the question of how the discount data should be stored. However, IKEA is proposed to continue to use the clear winner PostgreSQL as a DBMS. (Less)
Please use this url to cite or link to this publication:
author
Sinander, Annelie LU and Galonja, Alexandra LU
supervisor
organization
alternative title
Lämpliga databastekniker för kundrabatter
course
EITL05 20221
year
type
M2 - Bachelor Degree
subject
keywords
Database Technologies, Discounts, SQL, NoSQL, Latency, Signal-to-noise-ratio
report number
LU/LTH-EIT 2022-877
language
English
id
9091692
date added to LUP
2022-06-23 14:22:51
date last changed
2022-06-23 14:22:51
@misc{9091692,
  abstract     = {{This thesis has been conducted in order to compare how four different database management systems (DBMSs) perform in a limited number of areas, in order to draw conclusions about which one works best for storing discount data at IKEA. The DBMSs being evaluated are MongoDB, MariaDB, Neo4j and PostgreSQL. To enable a comparison, a form of testing was performed - a benchmark. The benchmarking consists of setting up a database for each DBMS to be compared. Discount data is then stored in each database and the execution time is measured for each test database. To accumulate information about how the database is set up but also about the different discounts used by IKEA, two interviews were conducted where this information was able to emerge.

The results from the testing and the interviews showed that PostgreSQL, which is the DBMS used today, performed the best overall. This was the case for the licensing metric, for instance, where Postgres was the only DBMS that does not come with a licensing cost. The results from one of the DBMSs, MongoDB, only consisted of estimates and not exact measurements and therefore the conclusion is not completely unambiguous. Furthermore, Neo4j's results when it came to the latency and scalability were deviating compared to the remaining DBMSs, which may have been due to an disadvantageously configured database.

Lastly, a number of development opportunities are presented which show that there is an immense potential to further explore the question of how the discount data should be stored. However, IKEA is proposed to continue to use the clear winner PostgreSQL as a DBMS.}},
  author       = {{Sinander, Annelie and Galonja, Alexandra}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Right fit Database Technology for Discounts}},
  year         = {{2022}},
}